基于在线自调整神经网络的最大功率跟踪方法研究  

RESEARCH ON MPPT METHOD BASED ON ONLINE SELF-ADJUSTING NEURAL NETWORK

作　　者：韩丽[1] 尚仪[1] 史丽萍[1] 

机构地区：[1]中国矿业大学信息与电气工程学院,徐州221116

出　　处：《太阳能学报》2015年第8期1848-1853,共6页Acta Energiae Solaris Sinica

基　　金：教育部博士点基金(20110095110014)

摘　　要：针对扰动观察法易在最大功率点附近振荡的问题,研究基于在线自调整神经网络和扰动观察法结合的最大功率跟踪策略。首先提出可在线调整结构的快速资源优化网络,然后以温度、光伏阵列电压和功率为网络输入,以最大功率点电压为输出,建立MPPT神经网络模型。若当前工作状态与神经网络模型记忆模式距离较近,则以神经网络输出U_N为U_(MPP)。否则,启用扰动观察法在U_N基础上精确定位最大功率点。同时将此运行状态数据作为样本训练调整网络结构,增加记忆模式,提高网络的输出精度。仿真实验证明本方法可避免系统在最大功率点的振荡。Aiming algorithm, at the oscillation phenomenon near the MPPT in conventional perturbation and observation tracking a maximum power point tracking algorithm combining online self-adjusting neural network and perturbation ＆ observation method was proposed. Firstly, rapid restructuring online resource optimization network was proposed, then MPPT neural network model with the temperature, PV array voltage and power for the input and voltage of maximum power point for the output was built. If the current operating state was close to memory mode of neural network model, then the neural network exported UN as UMPP. Otherwise, the perturbation and observation method was used to precisely position the maximum power point on the UN basis. The operating state data were used for the training sample to adjust network structure, add memory mode and improve output precision. The simulative results show that the method can avoid the oscillation near maximum power point.

关 键 词：光伏发电 最大功率点跟踪 在线自调整神经网络 扰动观察法 

分 类 号：TM615[电气工程—电力系统及自动化]